diff options
36 files changed, 385 insertions, 671 deletions
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index bac3fb406a74..30eb2457000c 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -1144,73 +1144,6 @@ struct ib_mr *ib_get_dma_mr(struct ib_pd *pd, int mr_access_flags) } EXPORT_SYMBOL(ib_get_dma_mr); -struct ib_mr *ib_reg_phys_mr(struct ib_pd *pd, - struct ib_phys_buf *phys_buf_array, - int num_phys_buf, - int mr_access_flags, - u64 *iova_start) -{ - struct ib_mr *mr; - int err; - - err = ib_check_mr_access(mr_access_flags); - if (err) - return ERR_PTR(err); - - if (!pd->device->reg_phys_mr) - return ERR_PTR(-ENOSYS); - - mr = pd->device->reg_phys_mr(pd, phys_buf_array, num_phys_buf, - mr_access_flags, iova_start); - - if (!IS_ERR(mr)) { - mr->device = pd->device; - mr->pd = pd; - mr->uobject = NULL; - atomic_inc(&pd->usecnt); - atomic_set(&mr->usecnt, 0); - } - - return mr; -} -EXPORT_SYMBOL(ib_reg_phys_mr); - -int ib_rereg_phys_mr(struct ib_mr *mr, - int mr_rereg_mask, - struct ib_pd *pd, - struct ib_phys_buf *phys_buf_array, - int num_phys_buf, - int mr_access_flags, - u64 *iova_start) -{ - struct ib_pd *old_pd; - int ret; - - ret = ib_check_mr_access(mr_access_flags); - if (ret) - return ret; - - if (!mr->device->rereg_phys_mr) - return -ENOSYS; - - if (atomic_read(&mr->usecnt)) - return -EBUSY; - - old_pd = mr->pd; - - ret = mr->device->rereg_phys_mr(mr, mr_rereg_mask, pd, - phys_buf_array, num_phys_buf, - mr_access_flags, iova_start); - - if (!ret && (mr_rereg_mask & IB_MR_REREG_PD)) { - atomic_dec(&old_pd->usecnt); - atomic_inc(&pd->usecnt); - } - - return ret; -} -EXPORT_SYMBOL(ib_rereg_phys_mr); - int ib_query_mr(struct ib_mr *mr, struct ib_mr_attr *mr_attr) { return mr->device->query_mr ? diff --git a/fs/nfs/blocklayout/blocklayout.h b/fs/nfs/blocklayout/blocklayout.h index 92dca9e90d8d..c556640dcf3b 100644 --- a/fs/nfs/blocklayout/blocklayout.h +++ b/fs/nfs/blocklayout/blocklayout.h @@ -46,13 +46,6 @@ struct pnfs_block_dev; -enum pnfs_block_volume_type { - PNFS_BLOCK_VOLUME_SIMPLE = 0, - PNFS_BLOCK_VOLUME_SLICE = 1, - PNFS_BLOCK_VOLUME_CONCAT = 2, - PNFS_BLOCK_VOLUME_STRIPE = 3, -}; - #define PNFS_BLOCK_MAX_UUIDS 4 #define PNFS_BLOCK_MAX_DEVICES 64 @@ -117,13 +110,6 @@ struct pnfs_block_dev { struct pnfs_block_dev_map *map); }; -enum exstate4 { - PNFS_BLOCK_READWRITE_DATA = 0, - PNFS_BLOCK_READ_DATA = 1, - PNFS_BLOCK_INVALID_DATA = 2, /* mapped, but data is invalid */ - PNFS_BLOCK_NONE_DATA = 3 /* unmapped, it's a hole */ -}; - /* sector_t fields are all in 512-byte sectors */ struct pnfs_block_extent { union { @@ -134,15 +120,12 @@ struct pnfs_block_extent { sector_t be_f_offset; /* the starting offset in the file */ sector_t be_length; /* the size of the extent */ sector_t be_v_offset; /* the starting offset in the volume */ - enum exstate4 be_state; /* the state of this extent */ + enum pnfs_block_extent_state be_state; /* the state of this extent */ #define EXTENT_WRITTEN 1 #define EXTENT_COMMITTING 2 unsigned int be_tag; }; -/* on the wire size of the extent */ -#define BL_EXTENT_SIZE (7 * sizeof(__be32) + NFS4_DEVICEID4_SIZE) - struct pnfs_block_layout { struct pnfs_layout_hdr bl_layout; struct rb_root bl_ext_rw; diff --git a/fs/nfs/blocklayout/dev.c b/fs/nfs/blocklayout/dev.c index e535599a0719..a861bbdfe577 100644 --- a/fs/nfs/blocklayout/dev.c +++ b/fs/nfs/blocklayout/dev.c @@ -22,7 +22,7 @@ bl_free_device(struct pnfs_block_dev *dev) kfree(dev->children); } else { if (dev->bdev) - blkdev_put(dev->bdev, FMODE_READ); + blkdev_put(dev->bdev, FMODE_READ | FMODE_WRITE); } } @@ -65,6 +65,11 @@ nfs4_block_decode_volume(struct xdr_stream *xdr, struct pnfs_block_volume *b) return -EIO; p = xdr_decode_hyper(p, &b->simple.sigs[i].offset); b->simple.sigs[i].sig_len = be32_to_cpup(p++); + if (b->simple.sigs[i].sig_len > PNFS_BLOCK_UUID_LEN) { + pr_info("signature too long: %d\n", + b->simple.sigs[i].sig_len); + return -EIO; + } p = xdr_inline_decode(xdr, b->simple.sigs[i].sig_len); if (!p) @@ -195,7 +200,7 @@ bl_parse_simple(struct nfs_server *server, struct pnfs_block_dev *d, if (!dev) return -EIO; - d->bdev = blkdev_get_by_dev(dev, FMODE_READ, NULL); + d->bdev = blkdev_get_by_dev(dev, FMODE_READ | FMODE_WRITE, NULL); if (IS_ERR(d->bdev)) { printk(KERN_WARNING "pNFS: failed to open device %d:%d (%ld)\n", MAJOR(dev), MINOR(dev), PTR_ERR(d->bdev)); diff --git a/fs/nfs/blocklayout/extent_tree.c b/fs/nfs/blocklayout/extent_tree.c index 31d0b5e53dfd..c59a59c37f3d 100644 --- a/fs/nfs/blocklayout/extent_tree.c +++ b/fs/nfs/blocklayout/extent_tree.c @@ -462,6 +462,12 @@ out: return err; } +static size_t ext_tree_layoutupdate_size(size_t count) +{ + return sizeof(__be32) /* number of entries */ + + PNFS_BLOCK_EXTENT_SIZE * count; +} + static void ext_tree_free_commitdata(struct nfs4_layoutcommit_args *arg, size_t buffer_size) { @@ -489,7 +495,7 @@ static int ext_tree_encode_commit(struct pnfs_block_layout *bl, __be32 *p, continue; (*count)++; - if (*count * BL_EXTENT_SIZE > buffer_size) { + if (ext_tree_layoutupdate_size(*count) > buffer_size) { /* keep counting.. */ ret = -ENOSPC; continue; @@ -530,7 +536,7 @@ retry: if (unlikely(ret)) { ext_tree_free_commitdata(arg, buffer_size); - buffer_size = sizeof(__be32) + BL_EXTENT_SIZE * count; + buffer_size = ext_tree_layoutupdate_size(count); count = 0; arg->layoutupdate_pages = @@ -549,17 +555,14 @@ retry: } *start_p = cpu_to_be32(count); - arg->layoutupdate_len = sizeof(__be32) + BL_EXTENT_SIZE * count; + arg->layoutupdate_len = ext_tree_layoutupdate_size(count); if (unlikely(arg->layoutupdate_pages != &arg->layoutupdate_page)) { - __be32 *p = start_p; + void *p = start_p, *end = p + arg->layoutupdate_len; int i = 0; - for (p = start_p; - p < start_p + arg->layoutupdate_len; - p += PAGE_SIZE) { + for ( ; p < end; p += PAGE_SIZE) arg->layoutupdate_pages[i++] = vmalloc_to_page(p); - } } dprintk("%s found %zu ranges\n", __func__, count); diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index 29e3c1b011b7..624bef79ba7c 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -554,7 +554,7 @@ __be32 nfs4_callback_recallslot(struct cb_recallslotargs *args, void *dummy, status = htonl(NFS4_OK); nfs41_set_target_slotid(fc_tbl, args->crsa_target_highest_slotid); - nfs41_server_notify_target_slotid_update(cps->clp); + nfs41_notify_server(cps->clp); out: dprintk("%s: exit with status = %d\n", __func__, ntohl(status)); return status; diff --git a/fs/nfs/client.c b/fs/nfs/client.c index 4a90c9bb3135..57c5a02f6213 100644 --- a/fs/nfs/client.c +++ b/fs/nfs/client.c @@ -20,6 +20,7 @@ #include <linux/stat.h> #include <linux/errno.h> #include <linux/unistd.h> +#include <linux/sunrpc/addr.h> #include <linux/sunrpc/clnt.h> #include <linux/sunrpc/stats.h> #include <linux/sunrpc/metrics.h> @@ -285,116 +286,6 @@ void nfs_put_client(struct nfs_client *clp) } EXPORT_SYMBOL_GPL(nfs_put_client); -#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) -/* - * Test if two ip6 socket addresses refer to the same socket by - * comparing relevant fields. The padding bytes specifically, are not - * compared. sin6_flowinfo is not compared because it only affects QoS - * and sin6_scope_id is only compared if the address is "link local" - * because "link local" addresses need only be unique to a specific - * link. Conversely, ordinary unicast addresses might have different - * sin6_scope_id. - * - * The caller should ensure both socket addresses are AF_INET6. - */ -static int nfs_sockaddr_match_ipaddr6(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - const struct sockaddr_in6 *sin1 = (const struct sockaddr_in6 *)sa1; - const struct sockaddr_in6 *sin2 = (const struct sockaddr_in6 *)sa2; - - if (!ipv6_addr_equal(&sin1->sin6_addr, &sin2->sin6_addr)) - return 0; - else if (ipv6_addr_type(&sin1->sin6_addr) & IPV6_ADDR_LINKLOCAL) - return sin1->sin6_scope_id == sin2->sin6_scope_id; - - return 1; -} -#else /* !defined(CONFIG_IPV6) && !defined(CONFIG_IPV6_MODULE) */ -static int nfs_sockaddr_match_ipaddr6(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - return 0; -} -#endif - -/* - * Test if two ip4 socket addresses refer to the same socket, by - * comparing relevant fields. The padding bytes specifically, are - * not compared. - * - * The caller should ensure both socket addresses are AF_INET. - */ -static int nfs_sockaddr_match_ipaddr4(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - const struct sockaddr_in *sin1 = (const struct sockaddr_in *)sa1; - const struct sockaddr_in *sin2 = (const struct sockaddr_in *)sa2; - - return sin1->sin_addr.s_addr == sin2->sin_addr.s_addr; -} - -static int nfs_sockaddr_cmp_ip6(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - const struct sockaddr_in6 *sin1 = (const struct sockaddr_in6 *)sa1; - const struct sockaddr_in6 *sin2 = (const struct sockaddr_in6 *)sa2; - - return nfs_sockaddr_match_ipaddr6(sa1, sa2) && - (sin1->sin6_port == sin2->sin6_port); -} - -static int nfs_sockaddr_cmp_ip4(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - const struct sockaddr_in *sin1 = (const struct sockaddr_in *)sa1; - const struct sockaddr_in *sin2 = (const struct sockaddr_in *)sa2; - - return nfs_sockaddr_match_ipaddr4(sa1, sa2) && - (sin1->sin_port == sin2->sin_port); -} - -#if defined(CONFIG_NFS_V4_1) -/* - * Test if two socket addresses represent the same actual socket, - * by comparing (only) relevant fields, excluding the port number. - */ -int nfs_sockaddr_match_ipaddr(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - if (sa1->sa_family != sa2->sa_family) - return 0; - - switch (sa1->sa_family) { - case AF_INET: - return nfs_sockaddr_match_ipaddr4(sa1, sa2); - case AF_INET6: - return nfs_sockaddr_match_ipaddr6(sa1, sa2); - } - return 0; -} -EXPORT_SYMBOL_GPL(nfs_sockaddr_match_ipaddr); -#endif /* CONFIG_NFS_V4_1 */ - -/* - * Test if two socket addresses represent the same actual socket, - * by comparing (only) relevant fields, including the port number. - */ -static int nfs_sockaddr_cmp(const struct sockaddr *sa1, - const struct sockaddr *sa2) -{ - if (sa1->sa_family != sa2->sa_family) - return 0; - - switch (sa1->sa_family) { - case AF_INET: - return nfs_sockaddr_cmp_ip4(sa1, sa2); - case AF_INET6: - return nfs_sockaddr_cmp_ip6(sa1, sa2); - } - return 0; -} - /* * Find an nfs_client on the list that matches the initialisation data * that is supplied. @@ -421,7 +312,7 @@ static struct nfs_client *nfs_match_client(const struct nfs_client_initdata *dat if (clp->cl_minorversion != data->minorversion) continue; /* Match the full socket address */ - if (!nfs_sockaddr_cmp(sap, clap)) + if (!rpc_cmp_addr_port(sap, clap)) continue; atomic_inc(&clp->cl_count); diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c index 547308a5ec6f..3d8e4ffa0a33 100644 --- a/fs/nfs/dir.c +++ b/fs/nfs/dir.c @@ -583,26 +583,19 @@ out_nopages: } static -void nfs_readdir_free_pagearray(struct page **pages, unsigned int npages) +void nfs_readdir_free_pages(struct page **pages, unsigned int npages) { unsigned int i; for (i = 0; i < npages; i++) put_page(pages[i]); } -static -void nfs_readdir_free_large_page(void *ptr, struct page **pages, - unsigned int npages) -{ - nfs_readdir_free_pagearray(pages, npages); -} - /* * nfs_readdir_large_page will allocate pages that must be freed with a call - * to nfs_readdir_free_large_page + * to nfs_readdir_free_pagearray */ static -int nfs_readdir_large_page(struct page **pages, unsigned int npages) +int nfs_readdir_alloc_pages(struct page **pages, unsigned int npages) { unsigned int i; @@ -615,7 +608,7 @@ int nfs_readdir_large_page(struct page **pages, unsigned int npages) return 0; out_freepages: - nfs_readdir_free_pagearray(pages, i); + nfs_readdir_free_pages(pages, i); return -ENOMEM; } @@ -623,7 +616,6 @@ static int nfs_readdir_xdr_to_array(nfs_readdir_descriptor_t *desc, struct page *page, struct inode *inode) { struct page *pages[NFS_MAX_READDIR_PAGES]; - void *pages_ptr = NULL; struct nfs_entry entry; struct file *file = desc->file; struct nfs_cache_array *array; @@ -653,7 +645,7 @@ int nfs_readdir_xdr_to_array(nfs_readdir_descriptor_t *desc, struct page *page, memset(array, 0, sizeof(struct nfs_cache_array)); array->eof_index = -1; - status = nfs_readdir_large_page(pages, array_size); + status = nfs_readdir_alloc_pages(pages, array_size); if (status < 0) goto out_release_array; do { @@ -671,7 +663,7 @@ int nfs_readdir_xdr_to_array(nfs_readdir_descriptor_t *desc, struct page *page, } } while (array->eof_index < 0); - nfs_readdir_free_large_page(pages_ptr, pages, array_size); + nfs_readdir_free_pages(pages, array_size); out_release_array: nfs_readdir_release_array(page); out_label_free: diff --git a/fs/nfs/file.c b/fs/nfs/file.c index cc4fa1ed61fc..7538a8582384 100644 --- a/fs/nfs/file.c +++ b/fs/nfs/file.c @@ -82,7 +82,8 @@ nfs_file_release(struct inode *inode, struct file *filp) dprintk("NFS: release(%pD2)\n", filp); nfs_inc_stats(inode, NFSIOS_VFSRELEASE); - return nfs_release(inode, filp); + nfs_file_clear_open_context(filp); + return 0; } EXPORT_SYMBOL_GPL(nfs_file_release); diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c index 2a93bec7e6dd..13fe64b4e259 100644 --- a/fs/nfs/flexfilelayout/flexfilelayout.c +++ b/fs/nfs/flexfilelayout/flexfilelayout.c @@ -730,8 +730,6 @@ ff_layout_pg_get_mirror_count_write(struct nfs_pageio_descriptor *pgio, return FF_LAYOUT_MIRROR_COUNT(pgio->pg_lseg); /* no lseg means that pnfs is not in use, so no mirroring here */ - pnfs_put_lseg(pgio->pg_lseg); - pgio->pg_lseg = NULL; nfs_pageio_reset_write_mds(pgio); return 1; } diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 0adc7d245b3d..382c8a46c078 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -888,7 +888,7 @@ struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_c return ctx; } -static void nfs_file_clear_open_context(struct file *filp) +void nfs_file_clear_open_context(struct file *filp) { struct nfs_open_context *ctx = nfs_file_open_context(filp); @@ -919,12 +919,6 @@ int nfs_open(struct inode *inode, struct file *filp) return 0; } -int nfs_release(struct inode *inode, struct file *filp) -{ - nfs_file_clear_open_context(filp); - return 0; -} - /* * This function is called whenever some part of NFS notices that * the cached attributes have to be refreshed. diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h index 1dad18105ed0..9ab3b1c21bb4 100644 --- a/fs/nfs/internal.h +++ b/fs/nfs/internal.h @@ -219,10 +219,6 @@ static inline void nfs_fs_proc_exit(void) } #endif -#ifdef CONFIG_NFS_V4_1 -int nfs_sockaddr_match_ipaddr(const struct sockaddr *, const struct sockaddr *); -#endif - /* callback_xdr.c */ extern struct svc_version nfs4_callback_version1; extern struct svc_version nfs4_callback_version4; diff --git a/fs/nfs/nfs3xdr.c b/fs/nfs/nfs3xdr.c index 9b04c2e6fffc..267126d32ec0 100644 --- a/fs/nfs/nfs3xdr.c +++ b/fs/nfs/nfs3xdr.c @@ -1103,6 +1103,7 @@ static void nfs3_xdr_enc_symlink3args(struct rpc_rqst *req, { encode_diropargs3(xdr, args->fromfh, args->fromname, args->fromlen); encode_symlinkdata3(xdr, args); + xdr->buf->flags |= XDRBUF_WRITE; } /* diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index ea3bee919a76..50cfc4ca7a02 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -405,9 +405,7 @@ int nfs40_discover_server_trunking(struct nfs_client *clp, int nfs41_discover_server_trunking(struct nfs_client *clp, struct nfs_client **, struct rpc_cred *); extern void nfs4_schedule_session_recovery(struct nfs4_session *, int); -extern void nfs41_server_notify_target_slotid_update(struct nfs_client *clp); -extern void nfs41_server_notify_highest_slotid_update(struct nfs_client *clp); - +extern void nfs41_notify_server(struct nfs_client *); #else static inline void nfs4_schedule_session_recovery(struct nfs4_session *session, int err) { diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c index 3aa6a9ba5113..223bedda64ae 100644 --- a/fs/nfs/nfs4client.c +++ b/fs/nfs/nfs4client.c @@ -729,10 +729,7 @@ static bool nfs4_cb_match_client(const struct sockaddr *addr, return false; /* Match only the IP address, not the port number */ - if (!nfs_sockaddr_match_ipaddr(addr, clap)) - return false; - - return true; + return rpc_cmp_addr(addr, clap); } /* diff --git a/fs/nfs/nfs4idmap.c b/fs/nfs/nfs4idmap.c index 535dfc69c628..2e4902203c35 100644 --- a/fs/nfs/nfs4idmap.c +++ b/fs/nfs/nfs4idmap.c @@ -184,7 +184,7 @@ static struct key_type key_type_id_resolver = { .read = user_read, }; -static int nfs_idmap_init_keyring(void) +int nfs_idmap_init(void) { struct cred *cred; struct key *keyring; @@ -230,7 +230,7 @@ failed_put_cred: return ret; } -static void nfs_idmap_quit_keyring(void) +void nfs_idmap_quit(void) { key_revoke(id_resolver_cache->thread_keyring); unregister_key_type(&key_type_id_resolver); @@ -492,16 +492,6 @@ nfs_idmap_delete(struct nfs_client *clp) kfree(idmap); } -int nfs_idmap_init(void) -{ - return nfs_idmap_init_keyring(); -} - -void nfs_idmap_quit(void) -{ - nfs_idmap_quit_keyring(); -} - static int nfs_idmap_prepare_message(char *desc, struct idmap *idmap, struct idmap_msg *im, struct rpc_pipe_msg *msg) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 15ee8bd99b61..f4e5816a77b0 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -586,7 +586,7 @@ out_unlock: spin_unlock(&tbl->slot_tbl_lock); res->sr_slot = NULL; if (send_new_highest_used_slotid) - nfs41_server_notify_highest_slotid_update(session->clp); + nfs41_notify_server(session->clp); } int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) @@ -8661,6 +8661,7 @@ static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = { .reboot_recovery_ops = &nfs41_reboot_recovery_ops, .nograce_recovery_ops = &nfs41_nograce_recovery_ops, .state_renewal_ops = &nfs41_state_renewal_ops, + .mig_recovery_ops = &nfs41_mig_recovery_ops, }; #endif diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index f2e2ad894461..da73bc443238 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -2152,23 +2152,13 @@ void nfs4_schedule_session_recovery(struct nfs4_session *session, int err) } EXPORT_SYMBOL_GPL(nfs4_schedule_session_recovery); -static void nfs41_ping_server(struct nfs_client *clp) +void nfs41_notify_server(struct nfs_client *clp) { /* Use CHECK_LEASE to ping the server with a SEQUENCE */ set_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state); nfs4_schedule_state_manager(clp); } -void nfs41_server_notify_target_slotid_update(struct nfs_client *clp) -{ - nfs41_ping_server(clp); -} - -void nfs41_server_notify_highest_slotid_update(struct nfs_client *clp) -{ - nfs41_ping_server(clp); -} - static void nfs4_reset_all_state(struct nfs_client *clp) { if (test_and_set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0) { diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 558cd65dbdb7..c42459e45f62 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1154,7 +1154,9 @@ static void encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg * case NF4LNK: p = reserve_space(xdr, 4); *p = cpu_to_be32(create->u.symlink.len); - xdr_write_pages(xdr, create->u.symlink.pages, 0, create->u.symlink.len); + xdr_write_pages(xdr, create->u.symlink.pages, 0, + create->u.symlink.len); + xdr->buf->flags |= XDRBUF_WRITE; break; case NF4BLK: case NF4CHR: diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index 4984bbe55ff1..7c5718ba625e 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -77,8 +77,8 @@ EXPORT_SYMBOL_GPL(nfs_pgheader_init); void nfs_set_pgio_error(struct nfs_pgio_header *hdr, int error, loff_t pos) { spin_lock(&hdr->lock); - if (pos < hdr->io_start + hdr->good_bytes) { - set_bit(NFS_IOHDR_ERROR, &hdr->flags); + if (!test_and_set_bit(NFS_IOHDR_ERROR, &hdr->flags) + || pos < hdr->io_start + hdr->good_bytes) { clear_bit(NFS_IOHDR_EOF, &hdr->flags); hdr->good_bytes = pos - hdr->io_start; hdr->error = error; diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c index cd3c0403101b..bbd407b6bc1f 100644 --- a/fs/nfs/pnfs_nfs.c +++ b/fs/nfs/pnfs_nfs.c @@ -373,26 +373,31 @@ same_sockaddr(struct sockaddr *addr1, struct sockaddr *addr2) return false; } +/* + * Checks if 'dsaddrs1' contains a subset of 'dsaddrs2'. If it does, + * declare a match. + */ static bool _same_data_server_addrs_locked(const struct list_head *dsaddrs1, const struct list_head *dsaddrs2) { struct nfs4_pnfs_ds_addr *da1, *da2; - - /* step through both lists, comparing as we go */ - for (da1 = list_first_entry(dsaddrs1, typeof(*da1), da_node), - da2 = list_first_entry(dsaddrs2, typeof(*da2), da_node); - da1 != NULL && da2 != NULL; - da1 = list_entry(da1->da_node.next, typeof(*da1), da_node), - da2 = list_entry(da2->da_node.next, typeof(*da2), da_node)) { - if (!same_sockaddr((struct sockaddr *)&da1->da_addr, - (struct sockaddr *)&da2->da_addr)) - return false; + struct sockaddr *sa1, *sa2; + bool match = false; + + list_for_each_entry(da1, dsaddrs1, da_node) { + sa1 = (struct sockaddr *)&da1->da_addr; + match = false; + list_for_each_entry(da2, dsaddrs2, da_node) { + sa2 = (struct sockaddr *)&da2->da_addr; + match = same_sockaddr(sa1, sa2); + if (match) + break; + } + if (!match) + break; } - if (da1 == NULL && da2 == NULL) - return true; - - return false; + return match; } /* diff --git a/fs/nfs/write.c b/fs/nfs/write.c index fdee9270ca15..388f48079c43 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1812,7 +1812,7 @@ out_mark_dirty: return res; } -static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_control *wbc) +int nfs_write_inode(struct inode *inode, struct writeback_control *wbc) { struct nfs_inode *nfsi = NFS_I(inode); int flags = FLUSH_SYNC; @@ -1847,11 +1847,6 @@ out_mark_dirty: __mark_inode_dirty(inode, I_DIRTY_DATASYNC); return ret; } - -int nfs_write_inode(struct inode *inode, struct writeback_control *wbc) -{ - return nfs_commit_unstable_pages(inode, wbc); -} EXPORT_SYMBOL_GPL(nfs_write_inode); /* diff --git a/fs/nfsd/blocklayoutxdr.c b/fs/nfsd/blocklayoutxdr.c index 9aa2796da90d..6d834dc9bbc8 100644 --- a/fs/nfsd/blocklayoutxdr.c +++ b/fs/nfsd/blocklayoutxdr.c @@ -101,7 +101,7 @@ nfsd4_block_decode_layoutupdate(__be32 *p, u32 len, struct iomap **iomapp, } nr_iomaps = be32_to_cpup(p++); - expected = sizeof(__be32) + nr_iomaps * NFS4_BLOCK_EXTENT_SIZE; + expected = sizeof(__be32) + nr_iomaps * PNFS_BLOCK_EXTENT_SIZE; if (len != expected) { dprintk("%s: extent array size mismatch: %u/%u\n", __func__, len, expected); diff --git a/fs/nfsd/blocklayoutxdr.h b/fs/nfsd/blocklayoutxdr.h index fdc79037c0e7..6de925fe8499 100644 --- a/fs/nfsd/blocklayoutxdr.h +++ b/fs/nfsd/blocklayoutxdr.h @@ -7,13 +7,6 @@ struct iomap; struct xdr_stream; -enum pnfs_block_extent_state { - PNFS_BLOCK_READWRITE_DATA = 0, - PNFS_BLOCK_READ_DATA = 1, - PNFS_BLOCK_INVALID_DATA = 2, - PNFS_BLOCK_NONE_DATA = 3, -}; - struct pnfs_block_extent { struct nfsd4_deviceid vol_id; u64 foff; @@ -21,14 +14,6 @@ struct pnfs_block_extent { u64 soff; enum pnfs_block_extent_state es; }; -#define NFS4_BLOCK_EXTENT_SIZE 44 - -enum pnfs_block_volume_type { - PNFS_BLOCK_VOLUME_SIMPLE = 0, - PNFS_BLOCK_VOLUME_SLICE = 1, - PNFS_BLOCK_VOLUME_CONCAT = 2, - PNFS_BLOCK_VOLUME_STRIPE = 3, -}; /* * Random upper cap for the uuid length to avoid unbounded allocation. diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index b8e72aad919c..00121f298269 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -547,6 +547,24 @@ enum pnfs_notify_deviceid_type4 { NOTIFY_DEVICEID4_DELETE = 1 << 2, }; +enum pnfs_block_volume_type { + PNFS_BLOCK_VOLUME_SIMPLE = 0, + PNFS_BLOCK_VOLUME_SLICE = 1, + PNFS_BLOCK_VOLUME_CONCAT = 2, + PNFS_BLOCK_VOLUME_STRIPE = 3, +}; + +enum pnfs_block_extent_state { + PNFS_BLOCK_READWRITE_DATA = 0, + PNFS_BLOCK_READ_DATA = 1, + PNFS_BLOCK_INVALID_DATA = 2, + PNFS_BLOCK_NONE_DATA = 3, +}; + +/* on the wire size of a block layout extent */ +#define PNFS_BLOCK_EXTENT_SIZE \ + (7 * sizeof(__be32) + NFS4_DEVICEID4_SIZE) + #define NFL4_UFLG_MASK 0x0000003F #define NFL4_UFLG_DENSE 0x00000001 #define NFL4_UFLG_COMMIT_THRU_MDS 0x00000002 diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 874b77228fb9..c0e961474a52 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -353,7 +353,6 @@ extern void nfs_access_add_cache(struct inode *, struct nfs_access_entry *); extern void nfs_access_set_mask(struct nfs_access_entry *, u32); extern int nfs_permission(struct inode *, int); extern int nfs_open(struct inode *, struct file *); -extern int nfs_release(struct inode *, struct file *); extern int nfs_attribute_timeout(struct inode *inode); extern int nfs_attribute_cache_expired(struct inode *inode); extern int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode); @@ -371,6 +370,7 @@ extern struct nfs_open_context *nfs_find_open_context(struct inode *inode, struc extern struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, fmode_t f_mode); extern void nfs_inode_attach_open_context(struct nfs_open_context *ctx); extern void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx); +extern void nfs_file_clear_open_context(struct file *flip); extern struct nfs_lock_context *nfs_get_lock_context(struct nfs_open_context *ctx); extern void nfs_put_lock_context(struct nfs_lock_context *l_ctx); extern u64 nfs_compat_user_ino64(u64 fileid); diff --git a/include/linux/sunrpc/addr.h b/include/linux/sunrpc/addr.h index 07d8e53bedfc..9bc3538d09a9 100644 --- a/include/linux/sunrpc/addr.h +++ b/include/linux/sunrpc/addr.h @@ -46,8 +46,8 @@ static inline void rpc_set_port(struct sockaddr *sap, #define IPV6_SCOPE_DELIMITER '%' #define IPV6_SCOPE_ID_LEN sizeof("%nnnnnnnnnn") -static inline bool __rpc_cmp_addr4(const struct sockaddr *sap1, - const struct sockaddr *sap2) +static inline bool rpc_cmp_addr4(const struct sockaddr *sap1, + const struct sockaddr *sap2) { const struct sockaddr_in *sin1 = (const struct sockaddr_in *)sap1; const struct sockaddr_in *sin2 = (const struct sockaddr_in *)sap2; @@ -67,8 +67,8 @@ static inline bool __rpc_copy_addr4(struct sockaddr *dst, } #if IS_ENABLED(CONFIG_IPV6) -static inline bool __rpc_cmp_addr6(const struct sockaddr *sap1, - const struct sockaddr *sap2) +static inline bool rpc_cmp_addr6(const struct sockaddr *sap1, + const struct sockaddr *sap2) { const struct sockaddr_in6 *sin1 = (const struct sockaddr_in6 *)sap1; const struct sockaddr_in6 *sin2 = (const struct sockaddr_in6 *)sap2; @@ -122,15 +122,28 @@ static inline bool rpc_cmp_addr(const struct sockaddr *sap1, if (sap1->sa_family == sap2->sa_family) { switch (sap1->sa_family) { case AF_INET: - return __rpc_cmp_addr4(sap1, sap2); + return rpc_cmp_addr4(sap1, sap2); case AF_INET6: - return __rpc_cmp_addr6(sap1, sap2); + return rpc_cmp_addr6(sap1, sap2); } } return false; } /** + * rpc_cmp_addr_port - compare the address and port number of two sockaddrs. + * @sap1: first sockaddr + * @sap2: second sockaddr + */ +static inline bool rpc_cmp_addr_port(const struct sockaddr *sap1, + const struct sockaddr *sap2) +{ + if (!rpc_cmp_addr(sap1, sap2)) + return false; + return rpc_get_port(sap1) == rpc_get_port(sap2); +} + +/** * rpc_copy_addr - copy the address portion of one sockaddr to another * @dst: destination sockaddr * @src: source sockaddr diff --git a/include/linux/sunrpc/xprtrdma.h b/include/linux/sunrpc/xprtrdma.h index b17613052cc3..b7b279b54504 100644 --- a/include/linux/sunrpc/xprtrdma.h +++ b/include/linux/sunrpc/xprtrdma.h @@ -49,7 +49,7 @@ * a single chunk type per message is supported currently. */ #define RPCRDMA_MIN_SLOT_TABLE (2U) -#define RPCRDMA_DEF_SLOT_TABLE (32U) +#define RPCRDMA_DEF_SLOT_TABLE (128U) #define RPCRDMA_MAX_SLOT_TABLE (256U) #define RPCRDMA_DEF_INLINE (1024) /* default inline max */ diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index b0f898e3b2e7..43c1cf01c84b 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -2760,52 +2760,6 @@ static inline void ib_dma_free_coherent(struct ib_device *dev, } /** - * ib_reg_phys_mr - Prepares a virtually addressed memory region for use - * by an HCA. - * @pd: The protection domain associated assigned to the registered region. - * @phys_buf_array: Specifies a list of physical buffers to use in the - * memory region. - * @num_phys_buf: Specifies the size of the phys_buf_array. - * @mr_access_flags: Specifies the memory access rights. - * @iova_start: The offset of the region's starting I/O virtual address. - */ -struct ib_mr *ib_reg_phys_mr(struct ib_pd *pd, - struct ib_phys_buf *phys_buf_array, - int num_phys_buf, - int mr_access_flags, - u64 *iova_start); - -/** - * ib_rereg_phys_mr - Modifies the attributes of an existing memory region. - * Conceptually, this call performs the functions deregister memory region - * followed by register physical memory region. Where possible, - * resources are reused instead of deallocated and reallocated. - * @mr: The memory region to modify. - * @mr_rereg_mask: A bit-mask used to indicate which of the following - * properties of the memory region are being modified. - * @pd: If %IB_MR_REREG_PD is set in mr_rereg_mask, this field specifies - * the new protection domain to associated with the memory region, - * otherwise, this parameter is ignored. - * @phys_buf_array: If %IB_MR_REREG_TRANS is set in mr_rereg_mask, this - * field specifies a list of physical buffers to use in the new - * translation, otherwise, this parameter is ignored. - * @num_phys_buf: If %IB_MR_REREG_TRANS is set in mr_rereg_mask, this - * field specifies the size of the phys_buf_array, otherwise, this - * parameter is ignored. - * @mr_access_flags: If %IB_MR_REREG_ACCESS is set in mr_rereg_mask, this - * field specifies the new memory access rights, otherwise, this - * parameter is ignored. - * @iova_start: The offset of the region's starting I/O virtual address. - */ -int ib_rereg_phys_mr(struct ib_mr *mr, - int mr_rereg_mask, - struct ib_pd *pd, - struct ib_phys_buf *phys_buf_array, - int num_phys_buf, - int mr_access_flags, - u64 *iova_start); - -/** * ib_query_mr - Retrieves information about a specific memory region. * @mr: The memory region to retrieve information about. * @mr_attr: The attributes of the specified memory region. diff --git a/net/sunrpc/xprtrdma/fmr_ops.c b/net/sunrpc/xprtrdma/fmr_ops.c index f1e8dafbd507..cb25c89da623 100644 --- a/net/sunrpc/xprtrdma/fmr_ops.c +++ b/net/sunrpc/xprtrdma/fmr_ops.c @@ -39,6 +39,25 @@ static int fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct rpcrdma_create_data_internal *cdata) { + struct ib_device_attr *devattr = &ia->ri_devattr; + struct ib_mr *mr; + + /* Obtain an lkey to use for the regbufs, which are + * protected from remote access. + */ + if (devattr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY) { + ia->ri_dma_lkey = ia->ri_device->local_dma_lkey; + } else { + mr = ib_get_dma_mr(ia->ri_pd, IB_ACCESS_LOCAL_WRITE); + if (IS_ERR(mr)) { + pr_err("%s: ib_get_dma_mr for failed with %lX\n", + __func__, PTR_ERR(mr)); + return -ENOMEM; + } + ia->ri_dma_lkey = ia->ri_dma_mr->lkey; + ia->ri_dma_mr = mr; + } + return 0; } diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c index 04ea914201b2..63f282e770b8 100644 --- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -189,6 +189,11 @@ frwr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct ib_device_attr *devattr = &ia->ri_devattr; int depth, delta; + /* Obtain an lkey to use for the regbufs, which are + * protected from remote access. + */ + ia->ri_dma_lkey = ia->ri_device->local_dma_lkey; + ia->ri_max_frmr_depth = min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, devattr->max_fast_reg_page_list_len); diff --git a/net/sunrpc/xprtrdma/physical_ops.c b/net/sunrpc/xprtrdma/physical_ops.c index 41985d07fdb7..72cf8b15bbb4 100644 --- a/net/sunrpc/xprtrdma/physical_ops.c +++ b/net/sunrpc/xprtrdma/physical_ops.c @@ -23,6 +23,29 @@ static int physical_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct rpcrdma_create_data_internal *cdata) { + struct ib_device_attr *devattr = &ia->ri_devattr; + struct ib_mr *mr; + + /* Obtain an rkey to use for RPC data payloads. + */ + mr = ib_get_dma_mr(ia->ri_pd, + IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE | + IB_ACCESS_REMOTE_READ); + if (IS_ERR(mr)) { + pr_err("%s: ib_get_dma_mr for failed with %lX\n", + __func__, PTR_ERR(mr)); + return -ENOMEM; + } + ia->ri_dma_mr = mr; + + /* Obtain an lkey to use for regbufs. + */ + if (devattr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY) + ia->ri_dma_lkey = ia->ri_device->local_dma_lkey; + else + ia->ri_dma_lkey = ia->ri_dma_mr->lkey; + return 0; } @@ -51,7 +74,7 @@ physical_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, struct rpcrdma_ia *ia = &r_xprt->rx_ia; rpcrdma_map_one(ia->ri_device, seg, rpcrdma_data_dir(writing)); - seg->mr_rkey = ia->ri_bind_mem->rkey; + seg->mr_rkey = ia->ri_dma_mr->rkey; seg->mr_base = seg->mr_dma; seg->mr_nsegs = 1; return 1; diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c index 84ea37daef36..bc8bd6577467 100644 --- a/net/sunrpc/xprtrdma/rpc_rdma.c +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -71,6 +71,67 @@ static const char transfertypes[][12] = { }; #endif +/* The client can send a request inline as long as the RPCRDMA header + * plus the RPC call fit under the transport's inline limit. If the + * combined call message size exceeds that limit, the client must use + * the read chunk list for this operation. + */ +static bool rpcrdma_args_inline(struct rpc_rqst *rqst) +{ + unsigned int callsize = RPCRDMA_HDRLEN_MIN + rqst->rq_snd_buf.len; + + return callsize <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst); +} + +/* The client can't know how large the actual reply will be. Thus it + * plans for the largest possible reply for that particular ULP + * operation. If the maximum combined reply message size exceeds that + * limit, the client must provide a write list or a reply chunk for + * this request. + */ +static bool rpcrdma_results_inline(struct rpc_rqst *rqst) +{ + unsigned int repsize = RPCRDMA_HDRLEN_MIN + rqst->rq_rcv_buf.buflen; + + return repsize <= RPCRDMA_INLINE_READ_THRESHOLD(rqst); +} + +static int +rpcrdma_tail_pullup(struct xdr_buf *buf) +{ + size_t tlen = buf->tail[0].iov_len; + size_t skip = tlen & 3; + + /* Do not include the tail if it is only an XDR pad */ + if (tlen < 4) + return 0; + + /* xdr_write_pages() adds a pad at the beginning of the tail + * if the content in "buf->pages" is unaligned. Force the + * tail's actual content to land at the next XDR position + * after the head instead. + */ + if (skip) { + unsigned char *src, *dst; + unsigned int count; + + src = buf->tail[0].iov_base; + dst = buf->head[0].iov_base; + dst += buf->head[0].iov_len; + + src += skip; + tlen -= skip; + + dprintk("RPC: %s: skip=%zu, memmove(%p, %p, %zu)\n", + __func__, skip, dst, src, tlen); + + for (count = tlen; count; count--) + *dst++ = *src++; + } + + return tlen; +} + /* * Chunk assembly from upper layer xdr_buf. * @@ -122,6 +183,10 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, if (len && n == nsegs) return -EIO; + /* When encoding the read list, the tail is always sent inline */ + if (type == rpcrdma_readch) + return n; + if (xdrbuf->tail[0].iov_len) { /* the rpcrdma protocol allows us to omit any trailing * xdr pad bytes, saving the server an RDMA operation. */ @@ -297,8 +362,7 @@ out: * pre-registered memory buffer for this request. For small amounts * of data, this is efficient. The cutoff value is tunable. */ -static int -rpcrdma_inline_pullup(struct rpc_rqst *rqst, int pad) +static void rpcrdma_inline_pullup(struct rpc_rqst *rqst) { int i, npages, curlen; int copy_len; @@ -310,16 +374,9 @@ rpcrdma_inline_pullup(struct rpc_rqst *rqst, int pad) destp = rqst->rq_svec[0].iov_base; curlen = rqst->rq_svec[0].iov_len; destp += curlen; - /* - * Do optional padding where it makes sense. Alignment of write - * payload can help the server, if our setting is accurate. - */ - pad -= (curlen + 36/*sizeof(struct rpcrdma_msg_padded)*/); - if (pad < 0 || rqst->rq_slen - curlen < RPCRDMA_INLINE_PAD_THRESH) - pad = 0; /* don't pad this request */ - dprintk("RPC: %s: pad %d destp 0x%p len %d hdrlen %d\n", - __func__, pad, destp, rqst->rq_slen, curlen); + dprintk("RPC: %s: destp 0x%p len %d hdrlen %d\n", + __func__, destp, rqst->rq_slen, curlen); copy_len = rqst->rq_snd_buf.page_len; @@ -355,7 +412,6 @@ rpcrdma_inline_pullup(struct rpc_rqst *rqst, int pad) page_base = 0; } /* header now contains entire send message */ - return pad; } /* @@ -380,7 +436,7 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); struct rpcrdma_req *req = rpcr_to_rdmar(rqst); char *base; - size_t rpclen, padlen; + size_t rpclen; ssize_t hdrlen; enum rpcrdma_chunktype rtype, wtype; struct rpcrdma_msg *headerp; @@ -402,28 +458,15 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) /* * Chunks needed for results? * + * o Read ops return data as write chunk(s), header as inline. * o If the expected result is under the inline threshold, all ops - * return as inline (but see later). + * return as inline. * o Large non-read ops return as a single reply chunk. - * o Large read ops return data as write chunk(s), header as inline. - * - * Note: the NFS code sending down multiple result segments implies - * the op is one of read, readdir[plus], readlink or NFSv4 getacl. - */ - - /* - * This code can handle read chunks, write chunks OR reply - * chunks -- only one type. If the request is too big to fit - * inline, then we will choose read chunks. If the request is - * a READ, then use write chunks to separate the file data - * into pages; otherwise use reply chunks. */ - if (rqst->rq_rcv_buf.buflen <= RPCRDMA_INLINE_READ_THRESHOLD(rqst)) - wtype = rpcrdma_noch; - else if (rqst->rq_rcv_buf.page_len == 0) - wtype = rpcrdma_replych; - else if (rqst->rq_rcv_buf.flags & XDRBUF_READ) + if (rqst->rq_rcv_buf.flags & XDRBUF_READ) wtype = rpcrdma_writech; + else if (rpcrdma_results_inline(rqst)) + wtype = rpcrdma_noch; else wtype = rpcrdma_replych; @@ -432,21 +475,25 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) * * o If the total request is under the inline threshold, all ops * are sent as inline. - * o Large non-write ops are sent with the entire message as a - * single read chunk (protocol 0-position special case). * o Large write ops transmit data as read chunk(s), header as * inline. + * o Large non-write ops are sent with the entire message as a + * single read chunk (protocol 0-position special case). * - * Note: the NFS code sending down multiple argument segments - * implies the op is a write. - * TBD check NFSv4 setacl + * This assumes that the upper layer does not present a request + * that both has a data payload, and whose non-data arguments + * by themselves are larger than the inline threshold. */ - if (rqst->rq_snd_buf.len <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst)) + if (rpcrdma_args_inline(rqst)) { rtype = rpcrdma_noch; - else if (rqst->rq_snd_buf.page_len == 0) - rtype = rpcrdma_areadch; - else + } else if (rqst->rq_snd_buf.flags & XDRBUF_WRITE) { rtype = rpcrdma_readch; + } else { + r_xprt->rx_stats.nomsg_call_count++; + headerp->rm_type = htonl(RDMA_NOMSG); + rtype = rpcrdma_areadch; + rpclen = 0; + } /* The following simplification is not true forever */ if (rtype != rpcrdma_noch && wtype == rpcrdma_replych) @@ -458,7 +505,6 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) } hdrlen = RPCRDMA_HDRLEN_MIN; - padlen = 0; /* * Pull up any extra send data into the preregistered buffer. @@ -467,45 +513,15 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) */ if (rtype == rpcrdma_noch) { - padlen = rpcrdma_inline_pullup(rqst, - RPCRDMA_INLINE_PAD_VALUE(rqst)); - - if (padlen) { - headerp->rm_type = rdma_msgp; - headerp->rm_body.rm_padded.rm_align = - cpu_to_be32(RPCRDMA_INLINE_PAD_VALUE(rqst)); - headerp->rm_body.rm_padded.rm_thresh = - cpu_to_be32(RPCRDMA_INLINE_PAD_THRESH); - headerp->rm_body.rm_padded.rm_pempty[0] = xdr_zero; - headerp->rm_body.rm_padded.rm_pempty[1] = xdr_zero; - headerp->rm_body.rm_padded.rm_pempty[2] = xdr_zero; - hdrlen += 2 * sizeof(u32); /* extra words in padhdr */ - if (wtype != rpcrdma_noch) { - dprintk("RPC: %s: invalid chunk list\n", - __func__); - return -EIO; - } - } else { - headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero; - headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero; - headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero; - /* new length after pullup */ - rpclen = rqst->rq_svec[0].iov_len; - /* - * Currently we try to not actually use read inline. - * Reply chunks have the desirable property that - * they land, packed, directly in the target buffers - * without headers, so they require no fixup. The - * additional RDMA Write op sends the same amount - * of data, streams on-the-wire and adds no overhead - * on receive. Therefore, we request a reply chunk - * for non-writes wherever feasible and efficient. - */ - if (wtype == rpcrdma_noch) - wtype = rpcrdma_replych; - } - } + rpcrdma_inline_pullup(rqst); + headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero; + headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero; + headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero; + /* new length after pullup */ + rpclen = rqst->rq_svec[0].iov_len; + } else if (rtype == rpcrdma_readch) + rpclen += rpcrdma_tail_pullup(&rqst->rq_snd_buf); if (rtype != rpcrdma_noch) { hdrlen = rpcrdma_create_chunks(rqst, &rqst->rq_snd_buf, headerp, rtype); @@ -518,9 +534,9 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) if (hdrlen < 0) return hdrlen; - dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd padlen %zd" + dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd" " headerp 0x%p base 0x%p lkey 0x%x\n", - __func__, transfertypes[wtype], hdrlen, rpclen, padlen, + __func__, transfertypes[wtype], hdrlen, rpclen, headerp, base, rdmab_lkey(req->rl_rdmabuf)); /* @@ -534,26 +550,15 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) req->rl_send_iov[0].length = hdrlen; req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf); + req->rl_niovs = 1; + if (rtype == rpcrdma_areadch) + return 0; + req->rl_send_iov[1].addr = rdmab_addr(req->rl_sendbuf); req->rl_send_iov[1].length = rpclen; req->rl_send_iov[1].lkey = rdmab_lkey(req->rl_sendbuf); req->rl_niovs = 2; - - if (padlen) { - struct rpcrdma_ep *ep = &r_xprt->rx_ep; - - req->rl_send_iov[2].addr = rdmab_addr(ep->rep_padbuf); - req->rl_send_iov[2].length = padlen; - req->rl_send_iov[2].lkey = rdmab_lkey(ep->rep_padbuf); - - req->rl_send_iov[3].addr = req->rl_send_iov[1].addr + rpclen; - req->rl_send_iov[3].length = rqst->rq_slen - rpclen; - req->rl_send_iov[3].lkey = rdmab_lkey(req->rl_sendbuf); - - req->rl_niovs = 4; - } - return 0; } diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c index 680f888a9ddd..64443eb754ad 100644 --- a/net/sunrpc/xprtrdma/transport.c +++ b/net/sunrpc/xprtrdma/transport.c @@ -175,10 +175,8 @@ xprt_rdma_format_addresses6(struct rpc_xprt *xprt, struct sockaddr *sap) } static void -xprt_rdma_format_addresses(struct rpc_xprt *xprt) +xprt_rdma_format_addresses(struct rpc_xprt *xprt, struct sockaddr *sap) { - struct sockaddr *sap = (struct sockaddr *) - &rpcx_to_rdmad(xprt).addr; char buf[128]; switch (sap->sa_family) { @@ -302,7 +300,7 @@ xprt_setup_rdma(struct xprt_create *args) struct rpc_xprt *xprt; struct rpcrdma_xprt *new_xprt; struct rpcrdma_ep *new_ep; - struct sockaddr_in *sin; + struct sockaddr *sap; int rc; if (args->addrlen > sizeof(xprt->addr)) { @@ -333,26 +331,20 @@ xprt_setup_rdma(struct xprt_create *args) * Set up RDMA-specific connect data. */ - /* Put server RDMA address in local cdata */ - memcpy(&cdata.addr, args->dstaddr, args->addrlen); + sap = (struct sockaddr *)&cdata.addr; + memcpy(sap, args->dstaddr, args->addrlen); /* Ensure xprt->addr holds valid server TCP (not RDMA) * address, for any side protocols which peek at it */ xprt->prot = IPPROTO_TCP; xprt->addrlen = args->addrlen; - memcpy(&xprt->addr, &cdata.addr, xprt->addrlen); + memcpy(&xprt->addr, sap, xprt->addrlen); - sin = (struct sockaddr_in *)&cdata.addr; - if (ntohs(sin->sin_port) != 0) + if (rpc_get_port(sap)) xprt_set_bound(xprt); - dprintk("RPC: %s: %pI4:%u\n", - __func__, &sin->sin_addr.s_addr, ntohs(sin->sin_port)); - - /* Set max requests */ cdata.max_requests = xprt->max_reqs; - /* Set some length limits */ cdata.rsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA write max */ cdata.wsize = RPCRDMA_MAX_SEGS * PAGE_SIZE; /* RDMA read max */ @@ -375,8 +367,7 @@ xprt_setup_rdma(struct xprt_create *args) new_xprt = rpcx_to_rdmax(xprt); - rc = rpcrdma_ia_open(new_xprt, (struct sockaddr *) &cdata.addr, - xprt_rdma_memreg_strategy); + rc = rpcrdma_ia_open(new_xprt, sap, xprt_rdma_memreg_strategy); if (rc) goto out1; @@ -409,7 +400,7 @@ xprt_setup_rdma(struct xprt_create *args) INIT_DELAYED_WORK(&new_xprt->rx_connect_worker, xprt_rdma_connect_worker); - xprt_rdma_format_addresses(xprt); + xprt_rdma_format_addresses(xprt, sap); xprt->max_payload = new_xprt->rx_ia.ri_ops->ro_maxpages(new_xprt); if (xprt->max_payload == 0) goto out4; @@ -420,6 +411,9 @@ xprt_setup_rdma(struct xprt_create *args) if (!try_module_get(THIS_MODULE)) goto out4; + dprintk("RPC: %s: %s:%s\n", __func__, + xprt->address_strings[RPC_DISPLAY_ADDR], + xprt->address_strings[RPC_DISPLAY_PORT]); return xprt; out4: @@ -653,31 +647,30 @@ static void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq) if (xprt_connected(xprt)) idle_time = (long)(jiffies - xprt->last_used) / HZ; - seq_printf(seq, - "\txprt:\trdma %u %lu %lu %lu %ld %lu %lu %lu %Lu %Lu " - "%lu %lu %lu %Lu %Lu %Lu %Lu %lu %lu %lu\n", - - 0, /* need a local port? */ - xprt->stat.bind_count, - xprt->stat.connect_count, - xprt->stat.connect_time, - idle_time, - xprt->stat.sends, - xprt->stat.recvs, - xprt->stat.bad_xids, - xprt->stat.req_u, - xprt->stat.bklog_u, - - r_xprt->rx_stats.read_chunk_count, - r_xprt->rx_stats.write_chunk_count, - r_xprt->rx_stats.reply_chunk_count, - r_xprt->rx_stats.total_rdma_request, - r_xprt->rx_stats.total_rdma_reply, - r_xprt->rx_stats.pullup_copy_count, - r_xprt->rx_stats.fixup_copy_count, - r_xprt->rx_stats.hardway_register_count, - r_xprt->rx_stats.failed_marshal_count, - r_xprt->rx_stats.bad_reply_count); + seq_puts(seq, "\txprt:\trdma "); + seq_printf(seq, "%u %lu %lu %lu %ld %lu %lu %lu %llu %llu ", + 0, /* need a local port? */ + xprt->stat.bind_count, + xprt->stat.connect_count, + xprt->stat.connect_time, + idle_time, + xprt->stat.sends, + xprt->stat.recvs, + xprt->stat.bad_xids, + xprt->stat.req_u, + xprt->stat.bklog_u); + seq_printf(seq, "%lu %lu %lu %llu %llu %llu %llu %lu %lu %lu %lu\n", + r_xprt->rx_stats.read_chunk_count, + r_xprt->rx_stats.write_chunk_count, + r_xprt->rx_stats.reply_chunk_count, + r_xprt->rx_stats.total_rdma_request, + r_xprt->rx_stats.total_rdma_reply, + r_xprt->rx_stats.pullup_copy_count, + r_xprt->rx_stats.fixup_copy_count, + r_xprt->rx_stats.hardway_register_count, + r_xprt->rx_stats.failed_marshal_count, + r_xprt->rx_stats.bad_reply_count, + r_xprt->rx_stats.nomsg_call_count); } static int diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index 891c4ede2c20..f73d7a71035c 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -52,6 +52,7 @@ #include <linux/prefetch.h> #include <linux/sunrpc/addr.h> #include <asm/bitops.h> +#include <linux/module.h> /* try_module_get()/module_put() */ #include "xprt_rdma.h" @@ -414,6 +415,14 @@ connected: return 0; } +static void rpcrdma_destroy_id(struct rdma_cm_id *id) +{ + if (id) { + module_put(id->device->owner); + rdma_destroy_id(id); + } +} + static struct rdma_cm_id * rpcrdma_create_id(struct rpcrdma_xprt *xprt, struct rpcrdma_ia *ia, struct sockaddr *addr) @@ -440,6 +449,17 @@ rpcrdma_create_id(struct rpcrdma_xprt *xprt, } wait_for_completion_interruptible_timeout(&ia->ri_done, msecs_to_jiffies(RDMA_RESOLVE_TIMEOUT) + 1); + + /* FIXME: + * Until xprtrdma supports DEVICE_REMOVAL, the provider must + * be pinned while there are active NFS/RDMA mounts to prevent + * hangs and crashes at umount time. + */ + if (!ia->ri_async_rc && !try_module_get(id->device->owner)) { + dprintk("RPC: %s: Failed to get device module\n", + __func__); + ia->ri_async_rc = -ENODEV; + } rc = ia->ri_async_rc; if (rc) goto out; @@ -449,16 +469,17 @@ rpcrdma_create_id(struct rpcrdma_xprt *xprt, if (rc) { dprintk("RPC: %s: rdma_resolve_route() failed %i\n", __func__, rc); - goto out; + goto put; } wait_for_completion_interruptible_timeout(&ia->ri_done, msecs_to_jiffies(RDMA_RESOLVE_TIMEOUT) + 1); rc = ia->ri_async_rc; if (rc) - goto out; + goto put; return id; - +put: + module_put(id->device->owner); out: rdma_destroy_id(id); return ERR_PTR(rc); @@ -493,9 +514,11 @@ rpcrdma_clean_cq(struct ib_cq *cq) int rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg) { - int rc, mem_priv; struct rpcrdma_ia *ia = &xprt->rx_ia; struct ib_device_attr *devattr = &ia->ri_devattr; + int rc; + + ia->ri_dma_mr = NULL; ia->ri_id = rpcrdma_create_id(xprt, ia, addr); if (IS_ERR(ia->ri_id)) { @@ -519,11 +542,6 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg) goto out3; } - if (devattr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY) { - ia->ri_have_dma_lkey = 1; - ia->ri_dma_lkey = ia->ri_device->local_dma_lkey; - } - if (memreg == RPCRDMA_FRMR) { /* Requires both frmr reg and local dma lkey */ if (((devattr->device_cap_flags & @@ -539,42 +557,19 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg) if (!ia->ri_device->alloc_fmr) { dprintk("RPC: %s: MTHCAFMR registration " "not supported by HCA\n", __func__); - memreg = RPCRDMA_ALLPHYSICAL; + goto out3; } } - /* - * Optionally obtain an underlying physical identity mapping in - * order to do a memory window-based bind. This base registration - * is protected from remote access - that is enabled only by binding - * for the specific bytes targeted during each RPC operation, and - * revoked after the corresponding completion similar to a storage - * adapter. - */ switch (memreg) { case RPCRDMA_FRMR: ia->ri_ops = &rpcrdma_frwr_memreg_ops; break; case RPCRDMA_ALLPHYSICAL: ia->ri_ops = &rpcrdma_physical_memreg_ops; - mem_priv = IB_ACCESS_LOCAL_WRITE | - IB_ACCESS_REMOTE_WRITE | - IB_ACCESS_REMOTE_READ; - goto register_setup; + break; case RPCRDMA_MTHCAFMR: ia->ri_ops = &rpcrdma_fmr_memreg_ops; - if (ia->ri_have_dma_lkey) - break; - mem_priv = IB_ACCESS_LOCAL_WRITE; - register_setup: - ia->ri_bind_mem = ib_get_dma_mr(ia->ri_pd, mem_priv); - if (IS_ERR(ia->ri_bind_mem)) { - printk(KERN_ALERT "%s: ib_get_dma_mr for " - "phys register failed with %lX\n", - __func__, PTR_ERR(ia->ri_bind_mem)); - rc = -ENOMEM; - goto out3; - } break; default: printk(KERN_ERR "RPC: Unsupported memory " @@ -592,7 +587,7 @@ out3: ib_dealloc_pd(ia->ri_pd); ia->ri_pd = NULL; out2: - rdma_destroy_id(ia->ri_id); + rpcrdma_destroy_id(ia->ri_id); ia->ri_id = NULL; out1: return rc; @@ -606,19 +601,11 @@ out1: void rpcrdma_ia_close(struct rpcrdma_ia *ia) { - int rc; - dprintk("RPC: %s: entering\n", __func__); - if (ia->ri_bind_mem != NULL) { - rc = ib_dereg_mr(ia->ri_bind_mem); - dprintk("RPC: %s: ib_dereg_mr returned %i\n", - __func__, rc); - } - if (ia->ri_id != NULL && !IS_ERR(ia->ri_id)) { if (ia->ri_id->qp) rdma_destroy_qp(ia->ri_id); - rdma_destroy_id(ia->ri_id); + rpcrdma_destroy_id(ia->ri_id); ia->ri_id = NULL; } @@ -639,6 +626,12 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, struct ib_cq_init_attr cq_attr = {}; int rc, err; + if (devattr->max_sge < RPCRDMA_MAX_IOVS) { + dprintk("RPC: %s: insufficient sge's available\n", + __func__); + return -ENOMEM; + } + /* check provider's send/recv wr limits */ if (cdata->max_requests > devattr->max_qp_wr) cdata->max_requests = devattr->max_qp_wr; @@ -651,21 +644,13 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, if (rc) return rc; ep->rep_attr.cap.max_recv_wr = cdata->max_requests; - ep->rep_attr.cap.max_send_sge = (cdata->padding ? 4 : 2); + ep->rep_attr.cap.max_send_sge = RPCRDMA_MAX_IOVS; ep->rep_attr.cap.max_recv_sge = 1; ep->rep_attr.cap.max_inline_data = 0; ep->rep_attr.sq_sig_type = IB_SIGNAL_REQ_WR; ep->rep_attr.qp_type = IB_QPT_RC; ep->rep_attr.port_num = ~0; - if (cdata->padding) { - ep->rep_padbuf = rpcrdma_alloc_regbuf(ia, cdata->padding, - GFP_KERNEL); - if (IS_ERR(ep->rep_padbuf)) - return PTR_ERR(ep->rep_padbuf); - } else - ep->rep_padbuf = NULL; - dprintk("RPC: %s: requested max: dtos: send %d recv %d; " "iovs: send %d recv %d\n", __func__, @@ -748,7 +733,8 @@ out2: dprintk("RPC: %s: ib_destroy_cq returned %i\n", __func__, err); out1: - rpcrdma_free_regbuf(ia, ep->rep_padbuf); + if (ia->ri_dma_mr) + ib_dereg_mr(ia->ri_dma_mr); return rc; } @@ -775,8 +761,6 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) ia->ri_id->qp = NULL; } - rpcrdma_free_regbuf(ia, ep->rep_padbuf); - rpcrdma_clean_cq(ep->rep_attr.recv_cq); rc = ib_destroy_cq(ep->rep_attr.recv_cq); if (rc) @@ -788,6 +772,12 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) if (rc) dprintk("RPC: %s: ib_destroy_cq returned %i\n", __func__, rc); + + if (ia->ri_dma_mr) { + rc = ib_dereg_mr(ia->ri_dma_mr); + dprintk("RPC: %s: ib_dereg_mr returned %i\n", + __func__, rc); + } } /* @@ -825,7 +815,7 @@ retry: if (ia->ri_device != id->device) { printk("RPC: %s: can't reconnect on " "different device!\n", __func__); - rdma_destroy_id(id); + rpcrdma_destroy_id(id); rc = -ENETUNREACH; goto out; } @@ -834,7 +824,7 @@ retry: if (rc) { dprintk("RPC: %s: rdma_create_qp failed %i\n", __func__, rc); - rdma_destroy_id(id); + rpcrdma_destroy_id(id); rc = -ENETUNREACH; goto out; } @@ -845,7 +835,7 @@ retry: write_unlock(&ia->ri_qplock); rdma_destroy_qp(old); - rdma_destroy_id(old); + rpcrdma_destroy_id(old); } else { dprintk("RPC: %s: connecting...\n", __func__); rc = rdma_create_qp(ia->ri_id, ia->ri_pd, &ep->rep_attr); @@ -1229,75 +1219,6 @@ rpcrdma_mapping_error(struct rpcrdma_mr_seg *seg) (unsigned long long)seg->mr_dma, seg->mr_dmalen); } -static int -rpcrdma_register_internal(struct rpcrdma_ia *ia, void *va, int len, - struct ib_mr **mrp, struct ib_sge *iov) -{ - struct ib_phys_buf ipb; - struct ib_mr *mr; - int rc; - - /* - * All memory passed here was kmalloc'ed, therefore phys-contiguous. - */ - iov->addr = ib_dma_map_single(ia->ri_device, - va, len, DMA_BIDIRECTIONAL); - if (ib_dma_mapping_error(ia->ri_device, iov->addr)) - return -ENOMEM; - - iov->length = len; - - if (ia->ri_have_dma_lkey) { - *mrp = NULL; - iov->lkey = ia->ri_dma_lkey; - return 0; - } else if (ia->ri_bind_mem != NULL) { - *mrp = NULL; - iov->lkey = ia->ri_bind_mem->lkey; - return 0; - } - - ipb.addr = iov->addr; - ipb.size = iov->length; - mr = ib_reg_phys_mr(ia->ri_pd, &ipb, 1, - IB_ACCESS_LOCAL_WRITE, &iov->addr); - - dprintk("RPC: %s: phys convert: 0x%llx " - "registered 0x%llx length %d\n", - __func__, (unsigned long long)ipb.addr, - (unsigned long long)iov->addr, len); - - if (IS_ERR(mr)) { - *mrp = NULL; - rc = PTR_ERR(mr); - dprintk("RPC: %s: failed with %i\n", __func__, rc); - } else { - *mrp = mr; - iov->lkey = mr->lkey; - rc = 0; - } - - return rc; -} - -static int -rpcrdma_deregister_internal(struct rpcrdma_ia *ia, - struct ib_mr *mr, struct ib_sge *iov) -{ - int rc; - - ib_dma_unmap_single(ia->ri_device, - iov->addr, iov->length, DMA_BIDIRECTIONAL); - - if (NULL == mr) - return 0; - - rc = ib_dereg_mr(mr); - if (rc) - dprintk("RPC: %s: ib_dereg_mr failed %i\n", __func__, rc); - return rc; -} - /** * rpcrdma_alloc_regbuf - kmalloc and register memory for SEND/RECV buffers * @ia: controlling rpcrdma_ia @@ -1317,26 +1238,29 @@ struct rpcrdma_regbuf * rpcrdma_alloc_regbuf(struct rpcrdma_ia *ia, size_t size, gfp_t flags) { struct rpcrdma_regbuf *rb; - int rc; + struct ib_sge *iov; - rc = -ENOMEM; rb = kmalloc(sizeof(*rb) + size, flags); if (rb == NULL) goto out; - rb->rg_size = size; - rb->rg_owner = NULL; - rc = rpcrdma_register_internal(ia, rb->rg_base, size, - &rb->rg_mr, &rb->rg_iov); - if (rc) + iov = &rb->rg_iov; + iov->addr = ib_dma_map_single(ia->ri_device, + (void *)rb->rg_base, size, + DMA_BIDIRECTIONAL); + if (ib_dma_mapping_error(ia->ri_device, iov->addr)) goto out_free; + iov->length = size; + iov->lkey = ia->ri_dma_lkey; + rb->rg_size = size; + rb->rg_owner = NULL; return rb; out_free: kfree(rb); out: - return ERR_PTR(rc); + return ERR_PTR(-ENOMEM); } /** @@ -1347,10 +1271,15 @@ out: void rpcrdma_free_regbuf(struct rpcrdma_ia *ia, struct rpcrdma_regbuf *rb) { - if (rb) { - rpcrdma_deregister_internal(ia, rb->rg_mr, &rb->rg_iov); - kfree(rb); - } + struct ib_sge *iov; + + if (!rb) + return; + + iov = &rb->rg_iov; + ib_dma_unmap_single(ia->ri_device, + iov->addr, iov->length, DMA_BIDIRECTIONAL); + kfree(rb); } /* @@ -1363,9 +1292,11 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct rpcrdma_req *req) { + struct ib_device *device = ia->ri_device; struct ib_send_wr send_wr, *send_wr_fail; struct rpcrdma_rep *rep = req->rl_reply; - int rc; + struct ib_sge *iov = req->rl_send_iov; + int i, rc; if (rep) { rc = rpcrdma_ep_post_recv(ia, ep, rep); @@ -1376,22 +1307,15 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia, send_wr.next = NULL; send_wr.wr_id = RPCRDMA_IGNORE_COMPLETION; - send_wr.sg_list = req->rl_send_iov; + send_wr.sg_list = iov; send_wr.num_sge = req->rl_niovs; send_wr.opcode = IB_WR_SEND; - if (send_wr.num_sge == 4) /* no need to sync any pad (constant) */ - ib_dma_sync_single_for_device(ia->ri_device, - req->rl_send_iov[3].addr, - req->rl_send_iov[3].length, - DMA_TO_DEVICE); - ib_dma_sync_single_for_device(ia->ri_device, - req->rl_send_iov[1].addr, - req->rl_send_iov[1].length, - DMA_TO_DEVICE); - ib_dma_sync_single_for_device(ia->ri_device, - req->rl_send_iov[0].addr, - req->rl_send_iov[0].length, - DMA_TO_DEVICE); + + for (i = 0; i < send_wr.num_sge; i++) + ib_dma_sync_single_for_device(device, iov[i].addr, + iov[i].length, DMA_TO_DEVICE); + dprintk("RPC: %s: posting %d s/g entries\n", + __func__, send_wr.num_sge); if (DECR_CQCOUNT(ep) > 0) send_wr.send_flags = 0; diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h index f49dd8b38122..d252457ff21a 100644 --- a/net/sunrpc/xprtrdma/xprt_rdma.h +++ b/net/sunrpc/xprtrdma/xprt_rdma.h @@ -65,9 +65,8 @@ struct rpcrdma_ia { struct ib_device *ri_device; struct rdma_cm_id *ri_id; struct ib_pd *ri_pd; - struct ib_mr *ri_bind_mem; + struct ib_mr *ri_dma_mr; u32 ri_dma_lkey; - int ri_have_dma_lkey; struct completion ri_done; int ri_async_rc; unsigned int ri_max_frmr_depth; @@ -89,7 +88,6 @@ struct rpcrdma_ep { int rep_connected; struct ib_qp_init_attr rep_attr; wait_queue_head_t rep_connect_wait; - struct rpcrdma_regbuf *rep_padbuf; struct rdma_conn_param rep_remote_cma; struct sockaddr_storage rep_remote_addr; struct delayed_work rep_connect_worker; @@ -119,7 +117,6 @@ struct rpcrdma_ep { struct rpcrdma_regbuf { size_t rg_size; struct rpcrdma_req *rg_owner; - struct ib_mr *rg_mr; struct ib_sge rg_iov; __be32 rg_base[0] __attribute__ ((aligned(256))); }; @@ -165,8 +162,7 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb) * struct rpcrdma_buffer. N is the max number of outstanding requests. */ -/* temporary static scatter/gather max */ -#define RPCRDMA_MAX_DATA_SEGS (64) /* max scatter/gather */ +#define RPCRDMA_MAX_DATA_SEGS ((1 * 1024 * 1024) / PAGE_SIZE) #define RPCRDMA_MAX_SEGS (RPCRDMA_MAX_DATA_SEGS + 2) /* head+tail = 2 */ struct rpcrdma_buffer; @@ -258,16 +254,18 @@ struct rpcrdma_mr_seg { /* chunk descriptors */ char *mr_offset; /* kva if no page, else offset */ }; +#define RPCRDMA_MAX_IOVS (2) + struct rpcrdma_req { - unsigned int rl_niovs; /* 0, 2 or 4 */ - unsigned int rl_nchunks; /* non-zero if chunks */ - unsigned int rl_connect_cookie; /* retry detection */ - struct rpcrdma_buffer *rl_buffer; /* home base for this structure */ + unsigned int rl_niovs; + unsigned int rl_nchunks; + unsigned int rl_connect_cookie; + struct rpcrdma_buffer *rl_buffer; struct rpcrdma_rep *rl_reply;/* holder for reply buffer */ - struct ib_sge rl_send_iov[4]; /* for active requests */ - struct rpcrdma_regbuf *rl_rdmabuf; - struct rpcrdma_regbuf *rl_sendbuf; - struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS]; + struct ib_sge rl_send_iov[RPCRDMA_MAX_IOVS]; + struct rpcrdma_regbuf *rl_rdmabuf; + struct rpcrdma_regbuf *rl_sendbuf; + struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS]; }; static inline struct rpcrdma_req * @@ -342,6 +340,7 @@ struct rpcrdma_stats { unsigned long hardway_register_count; unsigned long failed_marshal_count; unsigned long bad_reply_count; + unsigned long nomsg_call_count; }; /* diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index 0030376327b7..86ed778bb18b 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -2279,13 +2279,14 @@ static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task) WARN_ON_ONCE(!xprt_lock_connect(xprt, task, transport)); - /* Start by resetting any existing state */ - xs_reset_transport(transport); - - if (transport->sock != NULL && !RPC_IS_SOFTCONN(task)) { + if (transport->sock != NULL) { dprintk("RPC: xs_connect delayed xprt %p for %lu " "seconds\n", xprt, xprt->reestablish_timeout / HZ); + + /* Start by resetting any existing state */ + xs_reset_transport(transport); + queue_delayed_work(rpciod_workqueue, &transport->connect_worker, xprt->reestablish_timeout); |